SanDisk Sansa c200v2, m200v4, clipv1, clipv2, clip+, and fuzev2

[jago]

Hallo,
just tried the latest revision and did some test_disk runs on it.
At a first glance it looks really good. Seems it was really the voltage scaling causing these errors.
But i keep an eye on it. Thanks.

Edit:
Unfortunatly this was not the solution. Nothing has really changed. The same errors like before. (Yes, I formated the SD before it.)

[Hillshum]

I'm still getting uSD lockups, even without voltage scaling. If I put my 2gb SanDisk card into my e200v2 while it's running, it will lockup when I try to open the file browser. If I boot while the card is in, it hangs on the boot screen.

[kugel.]

I'm still getting uSD lockups, even without voltage scaling. If I put my 2gb SanDisk card into my e200v2 while it's running, it will lockup when I try to open the file browser. If I boot while the card is in, it hangs on the boot screen.

What was the last revision the card worked with?

If you post in this very development thread, then please don't just post bug reports or problems. Help development or stop posting here.

[FlynDice]

@Hillshum:

Have you tried reformatting the card?  I have had what sounds like the same problem and formatting the card made it work again.  This was pre voltage-scaling times i'm talking about now.

re: voltage scaling

I have been able to duplicate the condition described by MC2739 regarding the microsd card by taking the voltage down to 1.05v in my e280v2 but other than that I have no other progress to report but I'm still looking...

[unemongod]

I wonder if the voltage scaling for the CPU is affecting the uSD card voltage also.  That might explain why some cards seem more sensitive than others to the corruption.  Unfortunately, I am not yet familiar enough with this platform to confirm it, and I cannot afford to destroy my only fuze right now.  Maybe I can find a uSD interface adapter to put a VM on, anyone know of such a beast?


Also, I have noticed that the USB code does not differentiate between power and data connections, if I apply a charging voltage only, it reboots immediately into the Sansa FW.  I am also concerned that the fs is not being committed and closed cleanly when this happens.  Does rockbox use a sync or async filesystem?

[saratoga]

Also, I have noticed that the USB code does not differentiate between power and data connections,

Thats because we don't have a USB stack yet, so we don't have anyway to tell if a voltage on the USB lines is a device or a charger.

I am also concerned that the fs is not being committed and closed cleanly when this happens.  Does rockbox use a sync or async filesystem?

Rockbox uses FAT or FAT32 depending on volume size and how you've formatted it.  Why are you concerned thats its not closed correctly?

[funman]

Also, I have noticed that the USB code does not differentiate between power and data connection
 Does rockbox use a sync or async filesystem?

press center btn while connecting usb & rockbox will charge without rebooting

afaict rockbox uses sync filesystem but some storage i/o is multithreaded so the fs could be corrupted... in rare cases anyway

[Hillshum]

@Hillshum:

Have you tried reformatting the card?  I have had what sounds like the same problem and formatting the card made it work again.  This was pre voltage-scaling times i'm talking about now.

It works now. Sorry for forgetting to reformat and then bothering you all

[JdGordon]

* JdGordon is rather late to the party...

funman, am I correct that buttons in the same column cant be used together as combos? i.e home+up on the clip?

[FlynDice]

re:voltage scaling

Well, I've tracked it all the way down to bit 31 of the OCR register of the uSD card: Card power up status bit, line 279 of ata_sd_as3525.c.  I tried upping the max tries to 500 with no luck.

   "2) This bit is set to LOW if the card has not finished the power up routine."

So the card is not able to finish the power up routine with the processor core voltage at 1.05 and for some of us 1.10 it seems.  I cannot find anywhere in the datasheet where it would hint that CVDD would affect PVDD or IOVDD where I would assume we're drawing power for the SD card from.  There are settings for both of these peripheral voltages(AS3514_SYSTEM) but we leave them at their default(higher voltage) settings as far as I can tell.

I'll toil away a bit longer on this I guess.

Of course I'm learning all this as I go so please speak up if you've got a suggested direction to follow!

[funman]

@JdGordon : right

@FlynDice : if the card power up doesnt finish in N loops rockbox should just ignore the card no? (and not deadlock)

[kugel.]

@JdGordon : right

@FlynDice : if the card power up doesnt finish in N loops rockbox should just ignore the card no? (and not deadlock)

I rather think it should try harder.

[FlynDice]

@funman:

I think we have 2 separate problems with the way voltage scaling was being implemented.

  I think the deadlocks were being caused by the loop that was waiting for the voltage reading to be >= 1.20v.  I believe using adc_read in the loop fixes that problem.

  There seems to be an additional unrelated problem with the uSD card power up at the lower CVDD settings for some reason.  The card will init fine at boot-up when CVDD is at 1.20v, and also if you force a boosted state through the debug menu, CVDD also at 1.20.  But whenever CVDD is low the card fails to power up.

  This leaves rockbox in a state where it thinks it has a microsd installed but coughs when it tries to access the non powered up card.  The fix you put in this morning helps it fail more gracefully but I think the problem to be figured out here is how to get the card to power up correctly with CVDD at a lower setting.

[funman]

The OF seems to use 1.10V for CVDD (on Fuze and Clip at least) so there must be something it does and we are not doing.

Do you still get µSD cards not working with CVDD voltage less than 1.20V ?

If yes which error code does the SD driver reports?

[FlynDice]

My uSD card works fine at 1.10v but others were having what sounds like this same problem at 1.10v.  mc2739 described the problem quite well and it is just what I'm seeing when I take CVDD down to 1.05v.

sd_init_card() is returning a -2 now indicating a timeout on the card power up, also it is tripping the panicf down in the sd_thread.

EDIT:  I have tried goosing the voltage a bit on the uSD card and also adding extra delay between MCI_POWER_UP and MCI_POWER_ON with no change in the result.